Gmail taps machine learning for early phishing detection

It was just a few weeks ago that Google was on the receiving end of a fairly big phishing scam, one that targeted Google Docs users with real-looking emails. The scam was only live for a short while before Google squashed it, but the risk remains and anyone can fall victim. Here to help protect its users is Google and its security updates to Gmail. Among other things, the company is utilizing machine learning to help detect phishing attempts and, ultimately, keep users safe.

Google has already utilized machine learning to help keep phishing attempts at bay, and the entire process is largely successful: the company says that its machine learning technology helps keep more than 99.9-percent of phishing emails from showing up in your Gmail inbox. The company is now aiming to improve upon that number using a new dedicated machine learning model for early phishing detection.

With this, the technology will 'selectively delay' any emails to conduct what Google calls a 'rigorous phishing analysis,' helping keep Gmail users safe from such scams. This early detection technology will affect less than 0.05-percent of emails on average, says Google. The technology works alongside Google Safe Browsing, another machine learning technology, to identify and then flag any phishing URLs or links suspected of such scams.

If you do click on a link that Google finds suspicious, Gmail will offer up a click-time warning, which alerts the user to the potential security risk the link poses. For business users, Google says it is also now showing 'unintended external reply warnings' in Gmail, which gives users a warning if Gmail thinks you may not have meant to send that last email. This warning won't appear if you frequently message the contact, so it shouldn't be intrusive.

SOURCE: Google